Artificial Intelligence (AI) is going to be present in all aspects of life, work, and entertainment. It is possible that AI is even going to manage almost the entire planet at a certain point in the future.

Data centers are not immune to this reality. On the contrary. 

In the AI era, data centers need to quickly adapt to the computing power that the future is going to bring. Adding the capability of leveraging AI to reshape business models, make decisions, and improve customer experiences will become a key driving force.  

Thinking about the needs that the AI era will bring, Chinese telecommunications vendor Huawei has come a step forward with the world's first AI-powered data center switch especially built for the AI era. Presented and demonstrated at MWC Barcelona, the CloudEngine 16800 has micro-second latency, zero packet loss, and 100 percent throughput to double the AI computing power of enterprise networks. 

ccording to Leon Wang, General Manager of Huawei Data Center Network Domain, with the CloudEngine Series Huawei is helping the digital transformation of industry such as finance, Internet, and carrier customers.

With CloudEngine 16800 Huawei expects to "help customers accelerate intelligent transformation, achieve pervasive use of AI, and jointly build a fully connected and intelligent world," said Wang. 

Data center switches in the AI era 

With the CloudEngine 16800 Huawei defines three characteristics of data center switches in the AI era: 

• Embedded AI chip 

• 48-port 400GE line card per slot

• The capability to evolve to the autonomous driving network, innovatively incorporates AI technologies into data center switches

According to Huawei, the pervasive use of AI will help customers accelerate smart transformation.

RELATED: DATAVIEW VR: THE WORKSPACE OF THE FUTURE

The AI computing power is affected by the performance of data center networks. This is becoming a key bottleneck of the AI commercial process. With the industry expecting an increase of the annual volume of data worldwide from 10 zettabytes in 2018 to 180 zettabytes --that is 180 billion terabytes) in 2025 existing 100GE data center networks are not going to be able to handle the predicted data flood. 

AI computing power of data centers and the need to improve Operations and Maintenance (O&M) capability 

Currently, the AI computing power of data centers can only reach up to 50 percent on a traditional Ethernet; this is due to a packet loss rate of 1/1000. Traditional manual O&M methods are not going to meet requirements as the number of data center servers continues to increase. In addition, the computing network, storage network, and data network become converged. 

With this in mind, the need for the development of innovative technologies that can quickly be introduced and adopted is paramount in order to improve the smart O&M capability.

More information: https://interestingengineering.com/how-ai-and-huawei-are-transforming-the-data-center